Applescript third pane

[pjrobertson]

Fixes the stuff in #2071

Thanks @Sesquipedalian for reporting this and and also Mike Petonic for alerting me to this on the dev forums. Quite a big usability problem.

There is now an addition of a '3rd pane not-optional' choice for 3-pane actions.
That is, if you set the 'get argument count' part to return -3 then the 3rd pane is optional. If it returns 3 then the 3rd pane is not optional (and will display by default).

I used -3 and 3 as I thought it was kind of intuitive -3 meaning optional, 3 meaning not optional.
Open to other suggestions though

[pjrobertson]

@skurfer any idea why the testEquality test is failing?

[skurfer]

There is now an addition of a '3rd pane not-optional' choice for 3-pane actions.

I don’t know this code at all, but there’s a comment that says “1 = dObj only. 2 = dObj + iObj. 3 = indirect Optional”. So didn’t 2 already mean that the third pane was required?

Personally, I think I’d make 1 and 2 the only valid values, and have a separate thing you could put in the AppleScript that returns a BOOL to indicate whether or not the third pane is required. But maybe we should get opinions from people writing AppleScript actions. 😃

[skurfer]

any idea why the testEquality test is failing?

Probably the same reason it always has. combined1 are combined2 are literally the same object when you run the test the first time (and that really shouldn’t be possible). If you run the test again, it will work correctly, but of course Travis only does it once.

This might be moot soon. If XCTAssertEqualObjects and XCTAssertNotEqualObjects are checking hash before they check isEqual:, the test will need to be adjusted.

[Sesquipedalian]

I suspect that this PR isn't really going to help. The problem in #2071 is not a matter of controlling whether the third pane is optional or required (as @skurfer already noted, setting get argument count to either 2 or 3 already does that anyway). The problem is that the third pane is being populated with random junk for AppleScript actions. To fix #2071, the third pane should cleared whenever the user selects an AppleScript action in the second pane.

[pjrobertson]

The problem is that the third pane is being populated with random junk for AppleScript actions.

That's what this PR mainly fixed, I added the indirect option code (which Rob's pointed out isn't needed) as an addition. I'll just remove that 2nd commit then we'll be good... (done. Force pushed)

[Sesquipedalian]

Oh, okay. I guess I misunderstood your initial post. :)

[skurfer]

I'll just remove that 2nd commit then we'll be good...

Yeah, just looking at the code @Sesquipedalian, I think the stuff that ends up in the third pane should be greatly improved.

[pjrobertson]

(what I was trying to fix in that un-needed commit was a bug where the 3rd plane closes when you enter text mode, but I've found this is universal, not just with AppleScripts).

What this does for your first script (indirect optional, when you tab to the 3rd pane) is to cause the pane to open, go into text mode, close, then reopen again

[skurfer]

The way this is, won’t the third pane always have null as the first result? Don’t we want it to be null if no type was specified, otherwise, be an array of things matching that type? Or does the null get thrown out automatically when there are other choices?

[pjrobertson]

So [NSNull null] basically means that there is no selection in the 3rd pane. The results array is still populated (if there's a iObject type specified)

Basically, I didn't think we should make any assumptions about what object should be selected (e.g. should it be the highest ranking object of that type or what?) so I thought: best to leave it blank and let the user decide. He can search for it or press the ↓ arrow to scroll through the results.

My idea to choose a 'better' default object is a TODO, as can be seen here:
https://github.com/quicksilver/Quicksilver/pull/2232/files#diff-56ce0ea6b3cc1cdd1b7838ac51e344a3R296

[skurfer]

OK. Just knowing it was intentional is good enough for me.

[petonic]

Already said thanks on the dev forum, but THANKS again. I'll check it out on the next release of QS.

[pjrobertson]

You're welcome. Thanks to you for bringing the issue to our attention :)